LAMP Background
description
Transcript of LAMP Background
1
Localized Aviation Model Output Statistics Program (LAMP): Improvements to convective forecasts in
response to user feedback
Judy E. Ghirardelli
National Weather ServiceMeteorological Development Laboratory
October 11, 2011
“Friends/Partners in Aviation Weather” ForumNBAA Convention
Las Vegas, NV
LAMP Background• Statistical Guidance of sensible weather
Produced hourly, 25-h forecast period Valid at stations (airports) and on a grid
• Elements of interest to Aviation:• Winds (at stations)• Ceiling height (at stations and gridded)• Visibility (at stations and gridded)• Thunderstorms (at stations and gridded)
Visibility
Thunderstorms
Ceiling Height
2
How Did New LAMP Convection Guidance Evolve ?
• Existing Product: LAMP Lightning (LAMP ltg) Predictand: ≥ 1 Cloud-to-Ground (CTG) lightning strike
• Review of existing practices to verify convection products (ESRL) indicates radar refl. of ≥ 40 dBZ used as indicator of “convection” Problem: the verifying “truth” is not consistent with what LAMP lightning
was intended to forecast• FAA evaluation of operational LAMP ltg probabilities
Lacks spatial detail, skill, and sharpness especially beyond 6 hours • MDL decisions (June 2010)
Define convection predictand: radar ≥ 40 dBZ and/or ≥ 1 CTG lightning strikes
Add NAM MOS (to GFS MOS) convection probabilities as additional model input
3
Thunderstorm (current)
Convection (future)
4
New LAMP Convective Guidance
• Features: Defined from Cloud-to-Ground
(CTG) ltg GFS MOS 3-h thunderstorm
probability predictors 2-h period / 20-km gridboxes 1-h cycle; 3 – 25 h projections Other predictors
• Criticisms: Convection can occur without
CTG lightning Thunderstorm probabilities lack
sharpness
• Features: Defined from CTG ltg / ≥ 40 dBZ
radar reflectivity GFS & NAM MOS 2-h convective
probability predictors 2-h period / 20-km gridboxes 1-h cycle; 3 – 25 h projections Other predictors
• Solution: Convection can be indicated when
there is little or no lightning Convection probabilities exhibit
good sharpness
5
Convection Potential• Four convection potential categories
No, low, medium, and high Each category is defined objectively from a pre-determined
probability threshold Each probability threshold corresponds to a prescribed bias
criterion, where bias is
~ 2.7 = low potential ~ 1.1 = medium potential
(lightning ~ 1.2) ~ 0.4 = high potential
• Convection potential aids interpretation of probabilities with peak values < 100%
6
LAMP Lightning (LTG) vs Convection(CNV) Prob. Skill for 1800 UTC Cycle
Cool season Spring season
Summer season
Independent sampleOct 2009 – Oct 2010
7
LAMP Lightning vs Convection Probability Reliability and Sharpness
New LAMP Convective Guidance
8
August 27, 2011: 1800 UTC cycle, Hurricane Irene
LAMP Lightning LAMP Convection
New LAMP Convective Guidance
9
Convection Probabilities
Convection Potential
August 25, 2011: 1200 UTC cycle, 6-8 hour projection
10
Current LAMPThunderstorm
New LAMPConvection
11
12
Note that this is a 23-hour projection, and the LAMP convective probabilities are about 90% while the LAMP thunderstorm probabilities are about 30%.
13
14
May 26, 2011: 0000 UTC cycle: Chicago, Atlanta, New York all affectedConvective Guidance Examples
15
Convective Guidance ExamplesSeptember 07, 2011: 0000 UTC cycle
16
September 07, 2011: 0000 UTC cycle: 22-24 hour projectionConvective Guidance Examples
Future Work: Additional Products
17
Verification Graphics: overlay probabilities with marker indicating if convection was
observed
Text bulletins at stations: to support
prototype Gate Forecasts
Implementation Plans
• Convection products produced in real time since March 2011 24 cycles per day (not supported 24x7) Web Graphics at:
http://weather.gov/mdl/lamp/compare.php http://weather.gov/mdl/lamp/convection.php
GRIB2 files available at:http://www.mdl.nws.noaa.gov/~glmp/conv_grib/
• Implement on CCS parallel system before March 2012• Available in experimental NDGD March 2012• Transmit grids on SBN/NOAAPORT – planned FY12/13
• Contact: [email protected] 18